The UK is a world leader in pharmaceuticals. Five out of the world’s top 20 medicines have been developed in UK laboratories, and 12 out of the top 20 medicines, prescribed by the NHS, are British.
On average, it takes around 10 years and about £200m to develop a new medicine. However, this investment has paid off in the UK, with exports worth over £5,000 million last year.
Not only are exports healthy but, within 20 years, pharmaceutical innovations are expected to bring 10,000 fewer deaths from coronary heart disease, along with a saving of around £4 billion in healthcare in England and Wales.
These predictions were made by Peter Read, president of the Association of the British Pharmaceutical Industry, when he opened Interphex, the pharmaceutical exhibition, last year at Olympia.
The V words
One of the basic needs to be met when manufacturing drugs is verification and validation of computer systems software.
Some of the developments for validation have been led by the Food and Drug Administration (FDA) regulations in the USA. A similar body exists in the UK, called the Medicines Control Agency (MCA), which is run on user fees.
In the EC, Directive 85/65/EEC sets out the requirements that member states have to follow when authorising medicines. Applications to produce new drug and biological products have to be reviewed by these authorities.
Some companies, such as AMEC, use a project life cycle model, where a systematic set of procedures is used from original conception, through specification, design and operation of software.
Manufacturing deficiencies in process plant, including recipe handling, can result in approvals for new plants being delayed until corrected. For existing plants, just being on track is not enough, and companies have to keep up to date with the latest developments.
One of the most important aspects of computer software is the batch control system, led by ISA’s S88 standard, and IEC 1131 programing for PLCs, etc.
In Europe, the European Batch Forum is one of the leading developers of the S88 batch standard, which is being widely applied in drug manufacturing to meet the requirements of accurately documenting, as well as controlling, the production of pharmaceuticals.
Batch control has been driven by the needs for improved quality, traceability and, in particular, validation. However, payback in terms of higher yields and shorter run times can take time for the best return to be obtained from the higher skills and investment needed.
Looking briefly at software-based systems for batch processes, most of the leading suppliers have an S88 `aware’ package. Honeywell has a `flexible formulation’ package which is suited to handling processes with several different ingredients. And, Foxboro has RBatchII, which is particularly strong on handling both large numbers of sequences, for up to 150 processes.
Windows NT is also widely used by many suppliers. They include PID with OpenBatch, which offers users with small batch applications a lower price tag. Finally, Pro-log has ProBatch, an integrated recipe management system, based on a single but flexible platform. Many others offer excellent products, including Fisher Rosemount, ABB and Moore Products.
In fact, batch processing also now accounts for around 50% of chemical production, with multi-product and multi-stream plants taking over from the large continuous process units.
To handle the strong growth in pharmaceutical processes, part of the industry’s investment in new technology is in computer models for bioprocess control systems.
It is predicted that many of these advances will take account of the unpredictability and non-linearity of biological processes.
Neural networks, unlike statistical models, are inherently non-linear and unlimited in dimensionality – and therefore suited to these types of processes. They are also fast, which can lead to savings in product development
Gensym’s graphical object-orientated G2 system development environment has been used by many of the world’s leading pharmaceutical companies to build intelligent real time software systems.
One recent application of G2 was to identify critical respiration peaks in fermentation processes for Eli Lilly. The rule- and model-based system allowed the profiles to be detected in real time on average 70% faster than previously, resulting in higher yields and improved product consistency.
Eli Lilly also uses G2’s Diagnostic Assistant (GDA) to monitor incoming production data, compare this against established control charts, and then provide the operators with expert advice. Also, when alarms occur, GDA looks at the current process conditions and uses its knowledge base to suggest the optimum trouble shooting technique.
Other industrial neural network software that can be used for pharmaceutical and fine chemical plants is available from Neural Ware in Pittsburgh PA, Ward Systems in Frederick MD, Al Ware Systems from Cleveland OH, Neural Systems in Vancouver and Neural Computer Services in Southampton (UK).
Integration facilities and enterprise business software are also widely used in this industry. One of the leading suppliers is SAP, said to be the world’s fifth largest software house and its products can integrate the entire business – from order processing, market analyses and production to accounting.
Honeywill and Stein (H&S), a chemicals distributor for 40 specialists manufacturers, has implemented SAP’s client/server R/3 software at its facilities in Sutton. H&S is now on line with integrated software used for managing all aspects of distribution, logistics, materials management and finance.
Coopers and Lybrand was responsible for implementing the SAP system for H&S. This involved a substantial business process redesign to minimise the impact of software complexity.
C&L also provided support for the steep learning curve that in-house staff had to achieve.